kernels.scalarization_template.cl Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of neureka Show documentation

A platform independent tensor library written in Java.

The newest version!

/*
    The kernel define in this file is used to perform element wise
    operations on the data within nd-arrays / tensors where
    the second operand of the element wise operation is a scalar which
    is the same across all elements of the data array...
    This makes it somewhat similar to broadcasting, however
    the kernel can also be used when one of the given operand nd-array is filled
    with the same value across all data array entries...

    The kernel is a rather high level kernel because it take into account the
    entirety of an nd-configuration of a given nd-array / tensor.
    This means that the supplied nd-array might be a slice of another tensor
    which might also have steps/spread or is positioned by an offset within
    some data array...
    This kernel is built to support all of this for element wise operations.

*/
//======================================================================================================================

    // These are from the "utility.cl" file! The methods convert types of indices...
    void _cfg_of_cfg( __global int* cfg, int* prv_cfg, int rank );
    int _i_of_i( int i, int* cfg, int rank );

//======================================================================================================================

    __kernel void scalarization_template (
        __global #DATA_TYPE# *drn, __global int *drn_conf,
        __global #DATA_TYPE# *src1, __global int *src1_conf,
        #DATA_TYPE# value,
        int rank,
        int d
    ) {
        int prv_drn_cfg[32];
        _cfg_of_cfg(drn_conf, prv_drn_cfg, rank);
        int prv_src1_cfg[32];
        _cfg_of_cfg(src1_conf, prv_src1_cfg, rank);

        unsigned int i = get_global_id( 0 );
        if ( d < 0 ) {
//-==-//
            drn[_i_of_i(i, prv_drn_cfg, rank)] = src1[_i_of_i(i, prv_src1_cfg, rank)] + value;
//-==-//
        } else {
//-==-//
            drn[_i_of_i(i, prv_drn_cfg, rank)] = 1;
//-==-//
        }

    }

//======================================================================================================================